13 December 2017 5 4K Report

Hello!

In our experiment, we have 3 types of word pairs: (1) Unrelated, (2) Remotely related and (3) Closely related. The participant is asked to judge whether the two words in each pair (e.g., dog - bone) are semantically related ("Yes" or "No" response).

Our hypothesis is that a certain psychological condition will be associated with increased "Yes" responses for the Unrelated and Remotely related trials (whereas the condition has no effect on the closely related trials).

Normally, one could threat "Unrelated" as false alarms and both "Remote" and "Close" as two respective signals. However, since we a priori expect that both "Unrelated" and "Remote" would be higher, the standard d' parameter might not be suitable (i.e., if both %H and %FA go up).

Therefore, I was wondering whether there is some kind of parameter that is conceptually inverse to d', i.e., that will use signal ("Close") as reference to both "Unrelated" or "Remote".

An example of hit rates for two subjects:

Healthy: Unrelated = 10%, Remote = 30%, Close = 90%

Patient: Unrelated = 30%, Remote = 50%, Close = 90%

Is d' applicable if comparing Unrelated with Remote? And even if comparing Unrelated and Close, it does not say that its due to increased noise and not decreased hit rate, right?

Thank you for any suggestion / transformation / R package or reference,

best!

Martin

More Martin Marko's questions See All
Similar questions and discussions